Datasets:
| {"task_id": "sci_kinetics-L1", "family": "sci_kinetics", "domain": "natural science", "level": 1, "title": "Fit first-order kinetics", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/fit.json", "artifact_kind": "json", "example_instruction": "/app/run.csv follows a first-order decay over time. Find the rate constant k as minus the slope of an ordinary least-squares line of ln(concentration) against time in seconds, and the half-life ln(2)/k. run.csv has time in seconds and concentration in mol/L. Write /app/fit.json with the keys k_per_s (1/s) and half_life_s (seconds), for example {\"k_per_s\": 0.005, \"half_life_s\": 138.6}."} | |
| {"task_id": "sci_kinetics-L2", "family": "sci_kinetics", "domain": "natural science", "level": 2, "title": "Fit first-order kinetics", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/fit.json", "artifact_kind": "json", "example_instruction": "/app/run.csv follows a first-order decay over time. Find the rate constant k as minus the slope of an ordinary least-squares line of ln(concentration) against time in seconds, and the half-life ln(2)/k. Mind the units in run.csv, and read lab_notes.txt. Write /app/fit.json with the keys k_per_s (1/s) and half_life_s (seconds), for example {\"k_per_s\": 0.005, \"half_life_s\": 138.6}."} | |
| {"task_id": "sci_kinetics-L3", "family": "sci_kinetics", "domain": "natural science", "level": 3, "title": "Fit first-order kinetics", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/fit.json", "artifact_kind": "json", "example_instruction": "/app/run_295K.csv and /app/run_316K.csv are the same first-order decay at two temperatures. For each run, k is minus the slope of an ordinary least-squares line of ln(concentration) against time in seconds. Then find the activation energy from the Arrhenius equation using the two k values and R = 8.314 J/(mol K). Write /app/fit.json with the keys k_low_per_s and k_high_per_s (1/s, for the colder and the warmer run) and ea_kj_per_mol (kJ/mol), for example {\"k_low_per_s\": 0.002, \"k_high_per_s\": 0.006, \"ea_kj_per_mol\": 60.0}."} | |
| {"task_id": "sci_stoich-L1", "family": "sci_stoich", "domain": "natural science", "level": 1, "title": "Limiting reagent and yield", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/batch_report.json", "artifact_kind": "json", "example_instruction": "/app/reaction.txt describes one batch. Using only the atomic masses in /app/atomic_masses.txt, find the limiting reagent, the molar mass of NaNO3, and the grams of NaNO3 isolated. Write /app/batch_report.json with the keys limiting_reagent (its formula as written in the equation), product_molar_mass (g/mol) and product_grams, for example {\"limiting_reagent\": \"AgNO3\", \"product_molar_mass\": 100.09, \"product_grams\": 12.5}."} | |
| {"task_id": "sci_stoich-L2", "family": "sci_stoich", "domain": "natural science", "level": 2, "title": "Limiting reagent and yield", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/batch_report.json", "artifact_kind": "json", "example_instruction": "/app/reaction.txt describes one batch. Using only the atomic masses in /app/atomic_masses.txt, find the limiting reagent, the molar mass of Mg(NO3)2, and the grams of Mg(NO3)2 isolated. Write /app/batch_report.json with the keys limiting_reagent (its formula as written in the equation), product_molar_mass (g/mol) and product_grams, for example {\"limiting_reagent\": \"Mg(OH)2\", \"product_molar_mass\": 100.09, \"product_grams\": 12.5}."} | |
| {"task_id": "sci_stoich-L3", "family": "sci_stoich", "domain": "natural science", "level": 3, "title": "Limiting reagent and yield", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.833, "screen_full_solves": 5, "screen_non_json_endings": 1, "screen_budget_endings": 0, "screen_mean_steps": 3.33, "artifact_path": "/app/batch_report.json", "artifact_kind": "json", "example_instruction": "/app/reaction.txt describes one batch. Using only the atomic masses in /app/atomic_masses.txt, find the limiting reagent, the molar mass of NaCl, and the grams of NaCl isolated. Write /app/batch_report.json with the keys limiting_reagent (its formula as written in the equation), product_molar_mass (g/mol) and product_grams, for example {\"limiting_reagent\": \"BaCl2*2H2O\", \"product_molar_mass\": 100.09, \"product_grams\": 12.5}."} | |
| {"task_id": "sci_units-L1", "family": "sci_units", "domain": "natural science", "level": 1, "title": "Clinical lab unit harmonization", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/harmonized.csv", "artifact_kind": "csv", "example_instruction": "/app/labs.csv has lab results from several hospital systems in mixed units. Convert every value to its test's canonical unit using /app/conversions.txt (canonical = value x multiply_by; values already in the canonical unit stay). Write /app/harmonized.csv with the header sample_id,test,value: one row per sample and test, value in the canonical unit, for example S101,glucose,5.42."} | |
| {"task_id": "sci_units-L2", "family": "sci_units", "domain": "natural science", "level": 2, "title": "Clinical lab unit harmonization", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.5, "screen_full_solves": 3, "screen_non_json_endings": 2, "screen_budget_endings": 1, "screen_mean_steps": 3.0, "artifact_path": "/app/harmonized.csv", "artifact_kind": "csv", "example_instruction": "/app/labs.csv has lab results from several hospital systems in mixed units. Convert every value to its test's canonical unit using /app/conversions.txt (canonical = value x multiply_by; values already in the canonical unit stay). Unit spellings vary in case and form (mg/dl, umol/l, \u00b5mol/L ...). Write /app/harmonized.csv with the header sample_id,test,value: one row per sample and test, value in the canonical unit, for example S101,glucose,5.42."} | |
| {"task_id": "sci_units-L3", "family": "sci_units", "domain": "natural science", "level": 3, "title": "Clinical lab unit harmonization", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.5, "screen_full_solves": 3, "screen_non_json_endings": 3, "screen_budget_endings": 2, "screen_mean_steps": 3.5, "artifact_path": "/app/harmonized.csv", "artifact_kind": "csv", "example_instruction": "/app/labs.csv has lab results from several hospital systems in mixed units. Convert every value to its test's canonical unit using /app/conversions.txt (canonical = value x multiply_by; values already in the canonical unit stay). Unit spellings vary. Read /app/qc_notes.txt. Write /app/harmonized.csv with the header sample_id,test,value: one row per sample and test, value in the canonical unit, for example S101,glucose,5.42."} | |
| {"task_id": "sci_protein-L1", "family": "sci_protein", "domain": "natural science", "level": 1, "title": "Protein expression fold changes", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/top_hits.csv", "artifact_kind": "csv", "example_instruction": "/app/expression.csv has raw protein intensities, control (ctrl_*) and treated (trt_*) replicates. For each protein, log2_fc = log2(mean treated / mean control). Write /app/top_hits.csv with the header protein,log2_fc: the 5 proteins with the largest |log2_fc|, largest first, for example P12345,-2.41."} | |
| {"task_id": "sci_protein-L2", "family": "sci_protein", "domain": "natural science", "level": 2, "title": "Protein expression fold changes", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/top_hits.csv", "artifact_kind": "csv", "example_instruction": "/app/expression.csv has raw protein intensities, control (ctrl_*) and treated (trt_*) replicates. For each protein, log2_fc = log2(mean treated / mean control). Blank cells are missing: average the values present and skip a protein with fewer than 2 values in either group. Write /app/top_hits.csv with the header protein,log2_fc: the 5 proteins with the largest |log2_fc|, largest first, for example P12345,-2.41."} | |
| {"task_id": "sci_protein-L3", "family": "sci_protein", "domain": "natural science", "level": 3, "title": "Protein expression fold changes", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.833, "screen_full_solves": 5, "screen_non_json_endings": 2, "screen_budget_endings": 4, "screen_mean_steps": 3.0, "artifact_path": "/app/top_hits.csv", "artifact_kind": "csv", "example_instruction": "/app/expression.csv has raw protein intensities, three control and three treated runs. For each protein, log2_fc = log2(mean treated / mean control). Column conditions are in /app/samples.csv. Blank cells are missing: average the values present and skip a protein with fewer than 2 values in either group. Only keep proteins whose Welch t statistic on log2 intensities has |t| >= 4. Write /app/top_hits.csv with the header protein,log2_fc: the 4 proteins with the largest |log2_fc|, largest first, for example P12345,-2.41."} | |
| {"task_id": "sci_flood-L1", "family": "sci_flood", "domain": "natural science", "level": 1, "title": "Flood days at river gauges", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/flood_results.csv", "artifact_kind": "csv", "example_instruction": "/app/gauge.csv has daily maximum river stage per USGS station and /app/stations.csv the flood stage of each station. A flood day is a day whose stage is at or above the station's flood stage. Count flood days per station from 2025-03-24 to 2025-03-30 inclusive. Write /app/flood_results.csv with the header station_id,flood_days and one row per station with at least one flood day, for example 04001000,3."} | |
| {"task_id": "sci_flood-L2", "family": "sci_flood", "domain": "natural science", "level": 2, "title": "Flood days at river gauges", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.833, "screen_full_solves": 5, "screen_non_json_endings": 3, "screen_budget_endings": 2, "screen_mean_steps": 4.17, "artifact_path": "/app/flood_results.csv", "artifact_kind": "csv", "example_instruction": "/app/gauge.csv has daily maximum river stage per USGS station and /app/stations.csv the flood stage of each station. A flood day is a day whose stage is at or above the station's flood stage. Count flood days per station from 2025-03-23 to 2025-03-29 inclusive. A blank stage (with a code such as Ice or Eqp) is a missing reading, not a flood. Write /app/flood_results.csv with the header station_id,flood_days and one row per station with at least one flood day, for example 04001000,3."} | |
| {"task_id": "sci_flood-L3", "family": "sci_flood", "domain": "natural science", "level": 3, "title": "Flood days at river gauges", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/flood_results.csv", "artifact_kind": "csv", "example_instruction": "/app/gauge.csv has daily maximum river stage per USGS station and /app/stations.csv the flood stage of each station. A flood day is a day whose stage is at or above the station's flood stage. Count flood days per station from 2025-03-21 to 2025-03-27 inclusive. A blank stage is a missing reading. Flood stages are in the unit given per station (1 m = 3.28084 ft). Write /app/flood_results.csv with the header station_id,flood_days and one row per station with at least one flood day, for example 04001000,3."} | |
| {"task_id": "sci_trend-L1", "family": "sci_trend", "domain": "natural science", "level": 1, "title": "Lake warming trend", "in_submission": false, "screen_episodes": 6, "screen_mean_reward": 1.0, "screen_full_solves": 6, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": 3.83, "artifact_path": "/app/trend.json", "artifact_kind": "json", "example_instruction": "/app/water_temperature.csv has the annual mean lake surface temperature. Find the warming trend of the annual means against year: the ordinary least-squares slope and the Sen (median pairwise) slope, both in C per decade. Write /app/trend.json with the keys ols_slope_c_per_decade and sen_slope_c_per_decade, for example {\"ols_slope_c_per_decade\": 0.31, \"sen_slope_c_per_decade\": 0.29}."} | |
| {"task_id": "sci_trend-L2", "family": "sci_trend", "domain": "natural science", "level": 2, "title": "Lake warming trend", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/trend.json", "artifact_kind": "json", "example_instruction": "/app/water_temperature.csv has monthly lake surface temperatures. Average each year with all 12 months into an annual mean; drop incomplete years. Find the warming trend of the annual means against year: the ordinary least-squares slope and the Sen (median pairwise) slope, both in C per decade. Write /app/trend.json with the keys ols_slope_c_per_decade and sen_slope_c_per_decade, for example {\"ols_slope_c_per_decade\": 0.31, \"sen_slope_c_per_decade\": 0.29}."} | |
| {"task_id": "sci_trend-L3", "family": "sci_trend", "domain": "natural science", "level": 3, "title": "Lake warming trend", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.667, "screen_full_solves": 4, "screen_non_json_endings": 2, "screen_budget_endings": 1, "screen_mean_steps": 3.67, "artifact_path": "/app/trend.json", "artifact_kind": "json", "example_instruction": "/app/water_temperature.csv has monthly lake surface temperatures. Average each year with all 12 months into an annual mean; drop incomplete years. Find the warming trend of the annual means against year: the ordinary least-squares slope and the Sen (median pairwise) slope, both in C per decade. Also the Mann-Kendall Z (no tie correction, continuity-corrected: (S-1)/sd if S>0, (S+1)/sd if S<0) and significant = yes if |Z| > 1.96 else no. Write /app/trend.json with the keys ols_slope_c_per_decade, sen_slope_c_per_decade, mk_z and significant, for example {\"ols_slope_c_per_decade\": 0.31, \"sen_slope_c_per_decade\": 0.29, \"mk_z\": 2.4, \"significant\": \"yes\"}."} | |
| {"task_id": "sci_quake-L1", "family": "sci_quake", "domain": "natural science", "level": 1, "title": "Earthquake furthest from a plate boundary", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/answer.json", "artifact_kind": "json", "example_instruction": "/app/quakes.csv lists earthquakes inside one tectonic plate and /app/boundary.csv the vertices of its boundary. The distance of an event from the boundary is the great-circle (haversine) distance to the nearest boundary vertex, Earth radius 6371 km. Find the event furthest from the boundary. Write /app/answer.json with the keys id and distance_km, for example {\"id\": \"us70001234\", \"distance_km\": 1234.5}."} | |
| {"task_id": "sci_quake-L2", "family": "sci_quake", "domain": "natural science", "level": 2, "title": "Earthquake furthest from a plate boundary", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/answer.json", "artifact_kind": "json", "example_instruction": "/app/quakes.csv lists earthquakes inside one tectonic plate and /app/boundary.csv the vertices of its boundary. The distance of an event from the boundary is the great-circle (haversine) distance to the nearest boundary vertex, Earth radius 6371 km. Only consider events with magnitude 5.0 or more. Find the event furthest from the boundary. Write /app/answer.json with the keys id and distance_km, for example {\"id\": \"us70001234\", \"distance_km\": 1234.5}."} | |
| {"task_id": "sci_quake-L3", "family": "sci_quake", "domain": "natural science", "level": 3, "title": "Earthquake furthest from a plate boundary", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.333, "screen_full_solves": 2, "screen_non_json_endings": 4, "screen_budget_endings": 0, "screen_mean_steps": 3.33, "artifact_path": "/app/answer.json", "artifact_kind": "json", "example_instruction": "/app/quakes.csv lists earthquakes inside one tectonic plate and /app/boundary.csv the vertices of its boundary. The distance of an event from the boundary is the great-circle (haversine) distance to the nearest boundary vertex, Earth radius 6371 km. Only consider shallow events (depth 70 km or less) with magnitude 5.0 or more. Mind the column order in boundary.csv. Find the event furthest from the boundary. Write /app/answer.json with the keys id and distance_km, for example {\"id\": \"us70001234\", \"distance_km\": 1234.5}."} | |
| {"task_id": "sci_period-L1", "family": "sci_period", "domain": "natural science", "level": 1, "title": "Rotation period from a light curve", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.5, "screen_full_solves": 3, "screen_non_json_endings": 2, "screen_budget_endings": 1, "screen_mean_steps": 3.33, "artifact_path": "/app/period.json", "artifact_kind": "json", "example_instruction": "/app/tess_lc.csv is a TESS light curve of a spotted star whose brightness varies with its rotation. Find the rotation period as the period of the highest peak of a Lomb-Scargle periodogram searched between 0.5 and 15 days, on a fine frequency grid. Write /app/period.json with the key period_days, for example {\"period_days\": 3.42}. Within 1% is accepted."} | |
| {"task_id": "sci_period-L2", "family": "sci_period", "domain": "natural science", "level": 2, "title": "Rotation period from a light curve", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.5, "screen_full_solves": 3, "screen_non_json_endings": 2, "screen_budget_endings": 0, "screen_mean_steps": 5.83, "artifact_path": "/app/period.json", "artifact_kind": "json", "example_instruction": "/app/tess_lc.csv is a TESS light curve of a spotted star whose brightness varies with its rotation. Find the rotation period as the period of the highest peak of a Lomb-Scargle periodogram searched between 0.5 and 15 days, on a fine frequency grid. Use only rows with quality 0. Write /app/period.json with the key period_days, for example {\"period_days\": 3.42}. Within 1% is accepted."} | |
| {"task_id": "sci_period-L3", "family": "sci_period", "domain": "natural science", "level": 3, "title": "Rotation period from a light curve", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.333, "screen_full_solves": 2, "screen_non_json_endings": 5, "screen_budget_endings": 0, "screen_mean_steps": 3.33, "artifact_path": "/app/period.json", "artifact_kind": "json", "example_instruction": "/app/tess_lc.csv is a TESS light curve of a spotted star whose brightness varies with its rotation. Find the rotation period as the period of the highest peak of a Lomb-Scargle periodogram searched between 0.5 and 15 days, on a fine frequency grid. Use only rows with quality 0. The data are two sectors separated by a gap, each with its own raw flux scale: divide each sector's flux by its own median before searching. Write /app/period.json with the key period_days, for example {\"period_days\": 3.42}. Within 1% is accepted."} | |
| {"task_id": "sci_picking-L1", "family": "sci_picking", "domain": "natural science", "level": 1, "title": "Seismic P and S picks", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.5, "screen_full_solves": 3, "screen_non_json_endings": 2, "screen_budget_endings": 1, "screen_mean_steps": 5.17, "artifact_path": "/app/picks.json", "artifact_kind": "json", "example_instruction": "/app/trace.csv is one seismic trace sampled at 100 Hz (index, amplitude). Pick the P-wave arrival (the first arrival) as a sample index, for example with an STA/LTA detector. Write /app/picks.json with the key p_index, for example {\"p_index\": 812}. Within 50 samples is accepted."} | |
| {"task_id": "sci_picking-L2", "family": "sci_picking", "domain": "natural science", "level": 2, "title": "Seismic P and S picks", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/picks.json", "artifact_kind": "json", "example_instruction": "/app/trace.csv is one seismic trace sampled at 100 Hz (index, amplitude). Pick the P-wave arrival (the first arrival) and the S-wave arrival (the second, larger arrival) as a sample index, for example with an STA/LTA detector. Write /app/picks.json with the keys p_index and s_index, for example {\"p_index\": 812, \"s_index\": 1420}. Within 50 samples is accepted."} | |
| {"task_id": "sci_picking-L3", "family": "sci_picking", "domain": "natural science", "level": 3, "title": "Seismic P and S picks", "in_submission": true, "screen_episodes": 6, "screen_mean_reward": 0.167, "screen_full_solves": 1, "screen_non_json_endings": 4, "screen_budget_endings": 1, "screen_mean_steps": 4.5, "artifact_path": "/app/picks.csv", "artifact_kind": "csv", "example_instruction": "/app/traces/ has one seismic trace per station, sampled at 100 Hz (index, amplitude); the file name is the station. For each, pick the P-wave arrival (the first arrival) and the S-wave arrival (the second, larger arrival) as sample indices, for example with an STA/LTA detector. Write /app/picks.csv with the header station,p_index,s_index, for example ST12,812,1420. Within 50 samples is accepted."} | |
| {"task_id": "sci_vitals-L1", "family": "sci_vitals", "domain": "natural science", "level": 1, "title": "Breathing and heart rate from radar", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/vitals.json", "artifact_kind": "json", "example_instruction": "/app/chest.csv is 90 s of chest displacement from a 24 GHz continuous-wave radar (time in s, displacement in mm). Breathing is the dominant frequency between 0.1 and 0.6 Hz. Remove the mean and linear drift first. Write /app/vitals.json with the keys breaths_per_min (per minute), for example {\"breaths_per_min\": 15.2}. Within 1.5 per minute is accepted."} | |
| {"task_id": "sci_vitals-L2", "family": "sci_vitals", "domain": "natural science", "level": 2, "title": "Breathing and heart rate from radar", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/vitals.json", "artifact_kind": "json", "example_instruction": "/app/chest.csv is 90 s of chest displacement from a 24 GHz continuous-wave radar (time in s, displacement in mm). Breathing is the dominant frequency between 0.1 and 0.6 Hz; the heartbeat is the dominant frequency between 0.8 and 2.0 Hz. Remove the mean and linear drift first. Write /app/vitals.json with the keys breaths_per_min and heart_bpm (per minute), for example {\"breaths_per_min\": 15.2, \"heart_bpm\": 72.0}. Within 1.5 per minute is accepted."} | |
| {"task_id": "sci_vitals-L3", "family": "sci_vitals", "domain": "natural science", "level": 3, "title": "Breathing and heart rate from radar", "in_submission": false, "screen_episodes": 0, "screen_mean_reward": null, "screen_full_solves": 0, "screen_non_json_endings": 0, "screen_budget_endings": 0, "screen_mean_steps": null, "artifact_path": "/app/vitals.csv", "artifact_kind": "csv", "example_instruction": "/app/recordings/ has 90 s chest-displacement recordings from a 24 GHz radar, one file per recording; the first line gives its sample rate. Breathing is the dominant frequency between 0.1 and 0.6 Hz, the heartbeat the dominant frequency between 0.8 and 2.0 Hz; remove the mean and linear drift first. Write /app/vitals.csv with the header recording,breaths_per_min,heart_bpm (per minute), for example rec_001,15.2,72.0. Within 1.5 per minute is accepted."} | |